1A blocked Strait of Hormuz has put a July rate hike back on the board
Polymarket's July FOMC contract is the deepest thing on the board this week — about $92m of volume against $4.4m of liquidity, resolving on 29 July — and it prices no change at 75%, a 25bp hike at 24%, a 50bp hike at about 1%, and a cut at roughly zero. The July argument is hold-versus-hike. Easing is not on the ballot. The longer-dated contract says the same thing with more room: a rate hike landing somewhere in 2026 prices at 72% Yes, on about $4.6m of volume.
The mechanism is oil. Brent is near $100 with physical cargoes trading closer to $110, and the disruption behind that is priced as near-certain rather than merely likely: Polymarket puts the odds that Strait of Hormuz traffic returns to normal by 31 July at 1%, on about $20.6m of volume. Push the date out to 31 December and it becomes 54%, a coin flip. So the near term is locked and the medium term is genuinely unresolved, which is the awkward combination for anyone modelling an energy cost. Reuters described the transmission as oil battering bonds, with the 10-year near 4.7% and rising — that reaches equities as a discount-rate problem well before it reaches them as an earnings problem. The dollar is at a 40-year high against the yen with the ECB on hold, which is the same rate gap showing up in currencies.
Joseph Wang, on Fed Guy, argues the Fed should hike on 29 July, and reads the market's own odds at roughly 30% — close enough to Polymarket's 24% that the two are not really in dispute. His case is that three inflation drivers are running at once, oil and AI capex and tariffs, with the energy shock no longer transitory and the tariff wall rebuilt on what he calls sturdier legal ground. The week's administration signals put a dated example behind that third leg: the generic-drug import tariff is set at 0% for two years, then 100%, then 200% — a scheduled price increase that lands well outside the window this Fed is deciding on. Wang adds a political-calendar point too: 29 July sits about as far from the midterms as a meeting can, which is when a hike is cheapest to deliver.
He is not expecting the Iran escalation to run indefinitely. His read is that the US backs down eventually, as it did in March, but that the S&P could fall roughly another 10% first, toward a 200-day moving average near 7,000. The consequence is uncomfortable for both sides of the trade: the growth-scare path and the inflation path currently point at the same equity outcome, and only one of them ends in a cut.
2The AI selling hit the companies paying for the build-out and went around memory and storage
The Nasdaq fell 0.64% on Friday and the damage was not spread evenly. raised capex only marginally and was punished 4-5%, printing its first negative free cash flow on record — the cleanest instance yet of the market charging a company for AI spending rather than crediting it. sits near a one-year low going into its 29 July print. and were caught in the same mega-cap pullback. kept sliding after its own results, per Benzinga.
Memory, storage and optical went the other way, on a supply story of their own: +6%, +8%, +6% on optical and comms, up somewhere between 14% and 20% depending on the report, and +4% pre-market on an earnings beat. CNBC characterised the week as a rotation out of chips and into software, which the memory tape flatly contradicts — the selling was in the index and the mega-caps, and it did not reach the parts of the chip complex where supply is tight.
Wang reads the whole thing as the AI trade topping: the semiconductor index has rolled over, TSMC and ASML popped and then faded on good earnings, and hyperscaler debt markets are demanding a higher risk premium. Note where that lands him, because it cuts against the week's winners. He argues Chinese open-weight models could reduce RAM and GPU demand, which is bad for precisely the AI-bottleneck names that just rallied, and he reads Jensen Huang's first tweet — an endorsement of open-weight models — as the vendor seeing the same thing. therefore sits on both sides of his argument at once.
A separate overhang sits on the same mega-caps. The EU fined $15bn, $3bn and $2.5bn, with 's still pending, and President Trump's response was to attack the EU on their behalf. The tone is protective of the companies; the effect of the post is to restate the size of the bill. So the punishment this week landed on the buyers of compute rather than its sellers, and the one argument that would take memory down alongside them is structural — not something the tape has started pricing.
3The case against S&P 500 concentration is a thirty-year argument, not a timing call
Wes Gray, on Excess Returns, spent an hour making a case that resolves on no date at all, and was explicit about it: bubbles cannot be timed, and the only timing edge he will grant is momentum and trend. What high prices do tell you is expected return over the next twenty to thirty years, and today's imply a low one. His structural claim follows from that — the S&P 500's concentration is an oversized bet on large-cap quality and growth, factors with poor long-run expected returns, and a diversified factor investor would trim that exposure.
Two of his sub-arguments contradict the standard version of a well-known result, which is what makes them worth carrying. The small-cap premium, he argues, is a value effect rather than a size effect: hold value constant, comparing equal-weighted mid and large caps against small caps, and the returns match despite roughly a tenfold difference in liquidity. And book-to-market is negatively correlated with quality, because it buys companies that are cheap for the reason that they do not earn. He prefers earnings-to-price, which carries cheapness and quality in one number, and credits Graham with arriving there a century ago.
On AI he is not a sceptic — he calls it genuinely transformative and describes himself as a heavy Claude user — but he thinks prices are now "set to the moon, not Everest", and expects the value effect to reassert, on the view that the 2010-2020 growth run was partly a one-time re-rating rather than a repeatable earnings story. This is a philosophy interview with no dated catalysts in it, which is the honest limit of the claim: it can be right about thirty years while every tactical read above it is right about this week.
What the sources recommended
Two buys came out of the day, one from a video and one relayed from a bank, and neither is a clean read.
(Jack in the Box) comes from Ross Givens, updating a short-squeeze thesis he already holds rather than opening a new one. The setup: about 6.7m shares short, roughly 40% of the float and essentially unchanged after an initial 45% squeeze, plus Russell index inclusion forcing passive buying. The fundamental leg is a $500m refinancing that pushed maturities out to 2029 and took the near-term bankruptcy narrative off the table, an ex-Taco Bell chief executive in Mark King, Del Taco sold, and a stock at roughly 0.2x sales and 4x forward EBITDA on his figures. His model: a base case near $26 over 12-24 months against about $14 now, a bull case in the $40s, and a bear case near $4 if the turnaround fails, with early-August earnings the next catalyst. He discloses re-entering around $14 after being stopped out. Three things travel with the call and cannot be detached from it — the video is heavily promotional for a paid subscription service, the framing is explicitly meme-squeeze ("the next GameStop"), and by his own account the stock averages a 10% daily range.
(Applied Materials) was not a call made on any of the day's videos. TheStreet, carried by Yahoo Finance, reported Goldman Sachs doubling down with an upgrade. That is a relayed bank rating rather than a thesis stated on the record, so there is nothing in the day's material to inspect behind it.
The most bearish view of the day was not expressed in a single name at all. Wang put his caution into duration — long bonds, for the yield and as an equity-downside hedge — and hedged that in the same breath by conceding they trade poorly. The nearest thing to a sell on the page is a reported rotation rather than a recommendation: CNBC described money moving out of chips and into software, which is a description of a trade already being put on.
The hike question and the capex question resolve on the same day
29 July carries the FOMC decision and the earnings of both and — the July rate argument and the capex-punishment read landing within hours of each other, on two of the largest remaining spenders. Before that, Netanyahu meets Trump on Tuesday, a pairing that on Wang's read has usually preceded escalation rather than de-escalation, while the near-term Hormuz contract prices normalisation at 1% either way. 's earnings arrive in early August, which is when that thesis stops being about float and starts being about same-store sales.
